High-fidelity genome and prime editing enabled by the AI-designed openCRISPR-1.

Area of Science:

• Biochemistry
• Molecular Biology
• Bioengineering

Background:

• CRISPR-Cas9 systems revolutionized genome engineering.
• High-fidelity Cas9 variants improve specificity but can reduce activity or constrain guide RNA design.

Purpose of the Study:

• To evaluate OpenCRISPR-1, an AI-designed nuclease, for genome editing efficiency and specificity.
• To compare OpenCRISPR-1 with Cas9 and high-fidelity variants.
• To assess OpenCRISPR-based prime editors.

Main Methods:

• Head-to-head comparison of OpenCRISPR-1 and Cas9 in human cells using amplicon sequencing and multiplex Digenome-seq.
• Off-target validation via targeted sequencing in HEK293T cells, iPSCs, and MRC-5 fibroblasts.
• Comparison of OpenCRISPR-based prime editors (OpenCRISPR-PE2, OpenCRISPR-PE7) with existing prime editors.

Main Results:

• OpenCRISPR-1 demonstrated Cas9-level editing efficiency with up to a 553-fold reduction in off-target mutations compared to Cas9.
• OpenCRISPR-1 achieved specificity matching or surpassing high-fidelity Cas9 variants across diverse guide RNA formats.
• OpenCRISPR-based prime editors showed comparable editing efficiencies with improved specificity.

Conclusions:

• Generative AI-guided protein design effectively overcomes the specificity-efficiency trade-off in genome editing.
• OpenCRISPR-1 expands the genome editing toolkit for research and therapeutics.
• This work signifies a new era in rational protein design for genome engineering.